DMCA
On masking and swamping robustness of outlier identifiers for univariate data. Submitted (available at www.utdallas.edu/∼serfling (2012)
| Citations: | 1 - 1 self |
BibTeX
@MISC{Wang12onmasking,
author = {Shanshan Wang and Robert Serfling},
title = {On masking and swamping robustness of outlier identifiers for univariate data. Submitted (available at www.utdallas.edu/∼serfling},
year = {2012}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
In the wide-ranging scope of modern statistical data analysis, the identification of outliers remains an important and challenging task. One must assess the robustness of an outlier identification procedure against the possibilities of masking (an outlier is undetected) and swamping (a nonoutlier is classified as an outlier). General foundations and criteria for studying robustness of outlier detection procedures against masking and swamping were developed in Serfling and Wang (2012), in terms of masking breakdown point and swamping breakdown point. Here this framework is applied to evaluate these robustness measures in the case of univariate data for two leading outlier identifiers, scaled deviation outlyingness and centered rank outlyingness. The findings shed new light on the comparison of (Median, MAD) versus (trimmed mean, trimmed standard deviation) in defining scaled deviation outlyingness, and on the robustness of the boxplot, a leading descriptive tool.
Keyphrases
univariate data breakdown point identification procedure new light trimmed mean modern statistical data analysis general foundation descriptive tool wide-ranging scope detection procedure challenging task deviation outlyingness standard deviation robustness measure rank outlyingness scaled deviation outlyingness
